Wendy Erdheim-Poch from Elementor showed WordCamp US 2025 attendees exactly how AI and Playwright work together in real QA workflows — including when AI fails and what to do instead. Practical, honest, and immediately applicable.
Most AI talks at tech conferences follow a predictable arc: impressive demos, optimistic claims, vague takeaways. Wendy Erdheim-Poch’s session at WordCamp US 2025 was the opposite. Her presentation, “Catch Bugs Faster: No Hype, Just Results with AI & Playwright”, was grounded entirely in real experience, including the parts where AI made things worse. Having attended both the contributor day and the main conference day at WordCamp US 2025, I can say this session drew genuine attention from QA engineers, developers, and product managers alike — which is itself a sign that the content resonated beyond a single specialty.
About the Speaker
Wendy Erdheim-Poch is a QA professional at Elementor, with a strong focus on both manual testing and automation development. Her work sits at the intersection of two things that don’t always coexist well: sharp exploratory testing instincts and scalable, maintainable automated test suites. She is particularly focused on accessibility and deliberate testing strategy, with a consistent goal of making products more reliable, usable, and inclusive before users ever encounter a problem.
What the Talk Was Actually About
The premise was straightforward: AI tools can genuinely improve QA workflows, but only when you understand what they’re good at, what they’re bad at, and where they will quietly mislead you. Wendy covered all three with equal honesty.
On the useful side, she walked through three concrete areas where AI earns its place in a QA process:
- Generating test scenarios — AI can surface edge cases a human team might not think to write, especially for complex user flows with many variable states.
- Accelerating debugging — Feeding logs and error output to an AI model can significantly cut the time it takes to identify what actually caused a test failure.
- Supporting test code structure — AI can help write selectors and scaffold basic Playwright test structures, which is useful when a team is building out automation coverage quickly.
But the section that clearly landed hardest with the audience was the one on when not to use AI. Wendy was direct about cases where AI tools produced inaccurate results or created a false sense of coverage — situations where automated confidence masked real gaps in testing. That kind of honest calibration is rare in conference talks about AI, and it’s what made this one worth attending.
Playwright as the Foundation
A significant portion of the session was dedicated to the practical integration of AI with Playwright, Microsoft’s open-source framework for browser automation testing. Wendy demonstrated how AI assistance can help write more stable selectors, build comprehensive test scenarios, and interpret test run results more efficiently.
One of the more interesting examples involved accessibility testing: using AI to automatically check for WCAG compliance and surface specific improvement suggestions, rather than relying solely on manual review or single-purpose accessibility scanners.
Quality Is Everyone’s Responsibility
One of the central arguments in the talk was that product quality shouldn’t be siloed inside a QA team. When AI and automated testing lower the barrier to running meaningful quality checks, developers and product managers can participate in that process without needing deep testing expertise. That matters especially in WordPress development, where teams are often small and every contributor’s decisions affect the final product.
Watch the Full Session (English)
Timestamps for the full session:
- 00:10 — Workshop introduction
- 00:31 — The value of community connections
- 01:14 — Introduction to Wendy Erdheim-Poch
- 02:06 — Wendy’s QA journey
- 03:49 — A real story about AI testing gone wrong
- 05:19 — Quality as a shared team responsibility
- 06:05 — Where AI genuinely helps
- 08:16 — Understanding AI’s actual role in QA
- 09:30 — Introduction to Playwright
- 10:31 — Integrating AI with Playwright
- 12:27 — Structuring a testing approach
- 24:02 — Sarah’s testing experience (case example)
- 30:01 — When NOT to use AI
- 34:15 — Closing remarks and Q&A
Why This Session Stands Out
The WordPress ecosystem is not immune to AI hype. Plenty of tools are being marketed right now on the strength of the word “AI” alone, with limited transparency about where they actually help and where they introduce new risks. Wendy’s talk was a useful counterweight to that noise: concrete, honest, and built entirely on real QA work at a company that ships software used by millions of people.
If you work on WordPress products and care about quality at any level of the stack, this is a session worth watching in full.
